Distributed Places
message-router
spawn-node-with-place-at
spawn-node-supervise-place-at
spawn-remote-racket-node
create-place-node
supervise-place-at
supervise-process-at
supervise-thread-at
restart-every
every-seconds
after-seconds
connect-to-named-place
log-message
event-container<%>
register
spawned-process%
get-pid
new
place-socket-bridge%
get-sc-id
new
socket-connection%
new
node%
new
sync-events
remote-node%
new
get-first-place
get-first-place-channel
get-log-prefix
launch-place
remote-connect
send-exit
node-send-exit
node-get-first-place
distributed-place-wait
remote-connection%
new
set-on-channel!
place%
new
wait-for-die
connection%
new
respawn-and-fire%
new
after-seconds%
new
restarter%
new
distributed-launch-path
ssh-bin-path
racket-path
build-distributed-launch-path
spawn-node-at
spawn-nodes/  join
*channel-put
*channel-get
*channel?
send-new-place-channel-to-named-dest
mr-spawn-remote-node
mr-supervise-named-dynamic-place-at
mr-connect-to
start-message-router/  thread
port-no?
DEFAULT-ROUTER-PORT
named-place-typed-channel%
new
get
tc-get
write-flush
printf/  f
displayln/  f
1 Define Remote Server
define-remote-server
define-named-remote-server
define-state
define-rpc
define-cast
log-to-parent
2 Async Bidirectional Channels
make-async-bi-channel
async-bi-channel?
async-bi-channel-get
async-bi-channel-put
3 Distributed Places MPI
rmpi-comm
rmpi-id
rmpi-cnt
rmpi-send
rmpi-recv
rmpi-init
rmpi-broadcast
rmpi-reduce
rmpi-barrier
rmpi-allreduce
rmpi-partition
rmpi-build-default-config
rmpi-launch
rmpi-finish
5.92

Distributed Places

Kevin Tew

+See also Distributed Places in The Racket Guide.

 (require racket/place/distributed)
  package: distributed-places-lib

Distributed places support programs whose computation may span physical machines. The design relies on machine nodes that perform computation. The programmer configures a new distributed system using a declarative syntax and callbacks. A node begins life with one initial place: the message router. After a node has been configured, its message router is activated by calling the message-router function. The message router listens on a TCP port for incoming connections from other nodes in the distributed system. Places can be spawned within the node by sending place-spawn request messages to the node’s message router.

The distributed places implementation relies on two assumptions:

The following example illustrates a configuration and use of distributed places that starts a new node on the current machine and passes it a "Hello World" string:

Example:

> (module hello-world-example racket/base
    (require racket/place/distributed
             racket/place)
  
    (provide hello-world)
  
    (define (hello-world)
      (place ch
        (printf "hello-world received: ~a\n" (place-channel-get ch))
        (place-channel-put ch "Hello World\n")
        (printf "hello-world sent: Hello World\n")))
  
  
    (module+ main
      ; 1) spawns a node running at "localhost" and listenting on port
      ; 6344 for incomming connections.
      ; 2) connects to the node running at localhost:6344 and creates a
      ; place on that node by calling the hello-world procedure from
      ; the current module.
      ; 3) returns a remote-node% instance (node) and a
      ; remote-connection% instance (pl) for communicating with the
      ; new node and place
      (define-values (node pl)
        (spawn-node-supervise-place-at "localhost"
                                       #:listen-port 6344
                                       #:thunk #t
                                       (quote-module-path "..")
                                       'hello-world))
  
      ; starts a message router which adds three event-container<%>s to
      ; its list of events to handle: the node and two after-seconds
      ; event containers . Two seconds after the launch of the message-router, a
      ; message will be sent to the pl place.  After six seconds, the
      ; program and all spawned nodes and places will terminate.
      (message-router
        node
        (after-seconds 2
          (*channel-put pl "Hello")
          (printf "message-router received: ~a\n" (*channel-get pl)))
  
        (after-seconds 6
          (exit 0)))))

procedure

(message-router ec ...+)  void?

  ec : (is-a?/c event-container<%>)
Waits in an endless loop for one of many events to become ready. The message-router procedure constructs a node% instance to serve as the message router for the node. The message-router procedure then adds all the declared event-container<%>s to the node% and finally calls the never ending loop sync-events method, which handles events for the node.

procedure

(spawn-node-with-place-at 
  hostname 
  instance-module-path 
  instance-place-function-name 
  [#:listen-port port 
  #:initial-message initial-message 
  #:racket-path racket-path 
  #:ssh-bin-path ssh-path 
  #:distributed-launch-path launcher-path 
  #:restart-on-exit restart-on-exit 
  #:named place-name 
  #:thunk thunk]) 
  (is-a?/c remote-connection%)
  hostname : string?
  instance-module-path : module-path?
  instance-place-function-name : symbol?
  port : port-no? = DEFAULT-ROUTER-PORT
  initial-message : any = #f
  racket-path : string-path? = (racket-path)
  ssh-path : string-path? = (ssh-bin-path)
  launcher-path : string-path?
   = (path->string distributed-launch-path)
  restart-on-exit : any/c = #f
  place-name : (or/c #f symbol?) = #f
  thunk : (or/c #f #t) = #f
Spawns a new remote node at hostname with one instance place specified by the instance-module-path and instance-place-function-name.

When thunk is #f, the place is created as the result of the framework calling (dynamic-place instance-module-path instance-place-function-name). in the new node.

When thunk is #t the instance-place-function-name function should use dynamic-place or place to create and return an initial place in the new node.

When the place-name symbol is present a named place is created. The place-name symbol is used to establish later connections to the named place.

The result is a remote-node% instance, not a remote-connection%. Use get-first-place on the result to obtain a remote-connection%.

The restart-on-exit argument can be #t to instruct the remote-connection% instance to respawn the place on the remote node should it exit or terminate at any time. It can also be a procedure of zero arguments to implement the restart procedure, or it can be an object that support a restart method that takes a place argument.}

procedure

(spawn-node-supervise-place-at 
  hostname 
  instance-module-path 
  instance-place-function-name 
  [#:listen-port port 
  #:initial-message initial-message 
  #:racket-path racket-path 
  #:ssh-bin-path ssh-path 
  #:distributed-launch-path launcher-path 
  #:restart-on-exit restart-on-exit 
  #:named named 
  #:thunk thunk]) 
  
(is-a?/c remote-node%)
(is-a?/c remote-connection%)
  hostname : string?
  instance-module-path : module-path?
  instance-place-function-name : symbol?
  port : port-no? = DEFAULT-ROUTER-PORT
  initial-message : any = #f
  racket-path : string-path? = (racket-path)
  ssh-path : string-path? = (ssh-bin-path)
  launcher-path : string-path?
   = (path->string distributed-launch-path)
  restart-on-exit : any/c = #f
  named : (or/c #f string?) = #f
  thunk : (or/c #f #t) = #f
Like spawn-node-with-dynamic-place-at, but the result is two values: the new remote-node% and its remote-connection% instance.

procedure

(spawn-remote-racket-node 
  hostname 
  [#:listen-port port 
  #:racket-path racket-path 
  #:ssh-bin-path ssh-path 
  #:distributed-launch-path launcher-path] 
  #:use-current-ports use-current-ports) 
  (is-a?/c remote-node%)
  hostname : string?
  port : port-no? = DEFAULT-ROUTER-PORT
  racket-path : string-path? = (racket-path)
  ssh-path : string-path? = (ssh-bin-path)
  launcher-path : string-path?
   = (path->string distributed-launch-path)
  use-current-ports : #f
Spawns a new remote node at hostname and returns a remote-node% handle.

procedure

(create-place-node hostname 
  [#:listen-port port 
  #:racket-path racket-path 
  #:ssh-bin-path ssh-path 
  #:distributed-launch-path launcher-path 
  #:use-current-ports use-current-ports]) 
  (is-a?/c remote-node%)
  hostname : string?
  port : port-no? = DEFAULT-ROUTER-PORT
  racket-path : string-path? = (racket-path)
  ssh-path : string-path? = (ssh-bin-path)
  launcher-path : string-path?
   = (path->string distributed-launch-path)
  use-current-ports : boolean? = #t
Like spawn-remote-racket-node, but the current-output-port and current-error-port are used as the standard ports for the spawned process instead of new pipe ports.

procedure

(supervise-place-at remote-node 
  instance-module-path 
  instance-place-function-name 
  [#:restart-on-exit restart-on-exit 
  #:named named 
  #:thunk thunk]) 
  (is-a?/c remote-connection%)
  remote-node : (is-a?/c remote-node%)
  instance-module-path : module-path?
  instance-place-function-name : symbol?
  restart-on-exit : any/c = #f
  named : (or/c #f symbol?) = #f
  thunk : (or/c #f #t) = #f
When thunk is #f, creates a new place on remote-node by using dynamic-place to invoke instance-place-function-name from the module instance-module-path.

When thunk is #t, creates a new place at remote-node by executing the thunk exported as instance-place-function-name from the module instance-module-path. The function should use dynamic-place or place to create and return a place in the new node.

When the place-name symbol is present a named place is created. The place-name symbol is used to establish later connections to the named place.

procedure

(supervise-process-at hostname 
  commandline-argument ...+ 
  [#:listen-port port]) 
  (is-a?/c remote-process%)
  hostname : string?
  commandline-argument : string?
  port : port-no? = DEFAULT-ROUTER-PORT
Spawns an attached external process at host hostname.

procedure

(supervise-thread-at remote-node 
  instance-module-path 
  instance-thunk-function-name 
  [#:restart-on-exit restart-on-exit]) 
  (is-a?/c remote-connection%)
  remote-node : (is-a?/c remote-node%)
  instance-module-path : module-path?
  instance-thunk-function-name : symbol?
  restart-on-exit : any/c = #f
Creates a new threadon the remote-node by using dynamic-require to invoke instance-place-function-name from the module instance-module-path.

procedure

(restart-every seconds 
  [#:retry retry 
  #:on-final-fail on-final-fail]) 
  (is-a/c respawn-and-fire%)
  seconds : (number?)
  retry : (or/c number? #f) = #f
  on-final-fail : (or/c #f (-> any/c)) = #f
Returns a restarter% instance that should be supplied to a #:restart-on-exit argument.

syntax

(every-seconds seconds-expr body ....)

Returns a respawn-and-fire% instance that should be supplied to a message-router. The respawn-and-fire% instance executes bodys once every N seconds, where N is the result of seconds-expr.

syntax

(after-seconds seconds-expr body ....)

Returns a after-seconds% instance that should be supplied to a message-router. The after-seconds% instance executes the bodys after a delay of N seconds from the start of the event loop, where N is the result of seconds-expr.

procedure

(connect-to-named-place node name)

  (is-a?/c remote-connection%)
  node : (is-a?/c remote-node%)
  name : symbol?
Connects to a named place on the node named name and returns a remote-connection% object.

procedure

(log-message severity msg)  void?

  severity : (or/c 'fatal 'error 'warning 'info 'debug)
  msg : string?
Logs a message at the root node.

All objects that are supplied to the message-router must implement the event-container<%> interface. The message-router calls the register method on each supplied event-container<%> to obtain a list of events on which the event loop should wait.

method

(send an-event-container register events)  (listof events?)

  events : (listof events?)
Returns the list of events inside the event-container<%> that should be waited on by the message-router.

class

spawned-process% : class?

  superclass: object%

  extends: event-container<%>

method

(send a-spawned-process get-pid)  exact-positive-integer?

constructor

(new spawned-process% 
    [cmdline-list cmdline-list] 
    [[parent parent]]) 
  (is-a?/c spawned-process%)
  cmdline-list : (listof (or/c string? path?))
  parent : (is-a?/c remote-node%) = #f
The cmdline-list is a list of command line arguments of type string and/or path.

The parent argument is a remote-node% instance that will be notified when the process dies via a (send parent process-died this) call.

class

place-socket-bridge% : class?

  superclass: object%

  extends: event-container<%>

method

(send a-place-socket-bridge get-sc-id)

  exact-positive-integer?

constructor

(new place-socket-bridge% 
    [pch pch] 
    [sch sch] 
    [id id]) 
  (is-a?/c place-socket-bridge%)
  pch : place-channel?
  sch : (is-a?/c socket-connection%)
  id : exact-positive-integer?
The pch argument is a place-channel. Messages received on pch are forwarded to the socket-connection% sch via a dcgm message. e.g. (sconn-write-flush sch (dcgm DCGM-TYPE-INTER-DCHANNEL id id msg)) The id is a exact-positive-integer that identifies the socket-connection subchannel for this inter-node place connection.

class

socket-connection% : class?

  superclass: object%

  extends: event-container<%>

constructor

(new socket-connection% 
    [[host host] 
    [port port] 
    [retry-times retry-times] 
    [delay delay] 
    [background-connect? background-connect?] 
    [in in] 
    [out out] 
    [remote-node remote-node]]) 
  (is-a?/c socket-connection%)
  host : (or/c string? #f) = #f
  port : (or/c port-no? #f) = #f
  retry-times : exact-nonnegative-integer? = 30
  delay : number? = 1
  background-connect? : any/c = #f
  in : (or/c input-port? #f) = #f
  out : (or/c output-port #f) = #f
  remote-node : (or/c (is-a?/c remote-node%) #f) = #f
When a host and port are supplied a new tcp connection is established. If a input-port? and output-port? are supplied as in and out, the ports are used as a connection to the remote host. The retry-times argument specifies how many times to retry the connection attempt should it fail to connect and defaults to 30 retry attempts. Often a remote node is still booting up when a connection is attempted and the connection needs to be retried several times. The delay argument specifies how many seconds to wait between retry attempts. The background-connect? argument defaults to #t and specifies that the constructor should retry immediately and that connecion establishment should occur in the background. Finally, the remote-node argument specifies the remote-node% instance that should be notified should the connection fail.

class

node% : class?

  superclass: object%

  extends: event-container<%>
The node% instance controls a distributed places node. It launches places and routes inter-node place messages in the distributed system. The message-router form constructs a node% instance under the hood. Newly spawned nodes also have a node% instance in their initial place that serves as the node’s message router.

constructor

(new node% [[listen-port listen-port]])  (is-a?/c node%)

  listen-port : tcp-listen-port? = #f
Constructs a node% that will listen on listen-port for inter-node connections.

method

(send a-node sync-events)  void?

Starts the never ending event loop for this distributed places node.

class

remote-node% : class?

  superclass: object%

  extends: event-container<%>
Like node%, but for the remote API to a distributed places node. Instances of remote-node% are returned by create-place-node, spawn-remote-racket-node, and spawn-node-supervise-place-at.
A remote-node% is a place location in the sense of place-location?, which means that it can be supplied as the #:at argument to dynamic-place.

constructor

(new remote-node% 
    [[listen-port listen-port] 
    [restart-on-exit restart-on-exit]]) 
  (is-a?/c remote-node%)
  listen-port : tcp-listen-port? = #f
  restart-on-exit : any/c = #f
Constructs a node% that will listen on listen-port for inter-node connections.

When set to true the restart-on-exit parameter causes the specified node to be restarted when the ssh session spawning the node dies.

method

(send a-remote-node get-first-place)

  (is-a?/c remote-connection%)
Returns the remote-connection% object instance for the first place spawned on this node.

method

(send a-remote-node get-first-place-channel)  place-channel?

Returns the communication channel for the first place spawned on this node.

method

(send a-remote-node get-log-prefix)  string?

Returns (format "PLACE ~a:~a" host-name listen-port)

method

(send a-remote-node launch-place 
  place-exec 
  [#:restart-on-exit restart-on-exit 
  #:one-sided-place? one-sided-place?]) 
  (is-a?/c remote-connection%)
  place-exec : list?
  restart-on-exit : any/c = #f
  one-sided-place? : any/c = #f
Launches a place on the remote node represented by this remote-node% instance.

The place-exec argument describes how the remote place should be launched, and it should have one of the following shapes:
  • (list 'place place-module-path place-thunk)

  • (list 'dynamic-place place-module-path place-func)

The difference between these two launching methods is that the 'place version of place-exec expects a thunk to be exported by the module place-module-path. Executing the thunk is expected to create a new place and return a place descriptor to the newly created place. The 'dynamic-place version of place-exec expects place-func to be a function taking a single argument, the initial channel argument, and calls dynamic-place on behalf of the user and creates the new place from the place-module-path and place-func.

The restart-on-exit argument is treated in the same way as for spawn-node-with-dynamic-place-at.

The one-sided-place? argument is an internal use argument for launching remote places from within a place using the old design pattern.

method

(send a-remote-node remote-connect name)  remote-connection%

  name : string?
Connects to a named place on the remote node represented by this remote-node% instance.

method

(send a-remote-node send-exit)  void?

Sends a message instructing the remote node represented by this remote-node% instance to exit immediately

procedure

(node-send-exit remote-node%)  void?

  remote-node% : node
Sends node a message telling it to exit immediately.

procedure

(node-get-first-place remote-node%)

  (is-a?/c remote-connection%)
  remote-node% : node
Returns the remote-connection% instance of the first place spawned at this node

procedure

(distributed-place-wait remote-connection%)  void?

  remote-connection% : place
Waits for place to terminate.

class

remote-connection% : class?

  superclass: object%

  extends: event-container<%>
The remote-connection% instance provides a remote api to a place running on a remote distributed places node. It launches a places or connects to a named place and routes inter-node place messages to the remote place.

constructor

(new remote-connection% 
    [node node] 
    [place-exec place-exec] 
    [name name] 
    [restart-on-exit restart-on-exit] 
    [one-sided-place? one-sided-place?] 
    [on-channel on-channel]) 
  (is-a?/c remote-connection%)
  node : (is-a?/c remote-node%)
  place-exec : list?
  name : string?
  restart-on-exit : #f
  one-sided-place? : #f
  on-channel : #f
Constructs a remote-connection% instance.

The place-exec argument describes how the remote place should be launched in the same way as for launch-place in remote-node%.

The restart-on-exit argument is treated in the same way as for spawn-node-with-dynamic-place-at.

The one-sided-place? argument is an internal use argument for launching remote places from within a place using the old design pattern.

See set-on-channel! for description of on-channel argument.

method

(send a-remote-connection set-on-channel! callback)  void?

  callback : (-> channel msg void?)
Installs a handler function that handles messages from the remote place. The setup/distributed-docs module uses this callback to handle job completion messages.

class

place% : class?

  superclass: object%

  extends: event-container<%>
The place% instance represents a place launched on a distributed places node at that node. It launches a compute places and routes inter-node place messages to the place.

constructor

(new place%    
    [node node]    
    [place-exec place-exec]    
    [ch-id ch-id]    
    [sc sc]    
    [[on-place-dead on-place-dead]])  (is-a?/c place%)
  node : (is-a?/c remote-connection%)
  place-exec : list?
  ch-id : exact-positive-integer?
  sc : (is-a?/c socket-connection%)
  on-place-dead : (-> event void?) = default-on-place-dead
Constructs a remote-connection% instance. The place-exec argument describes how the remote place should be launched in the same way as for launch-place in remote-node%. The ch-id and sc arguments are internally used to establish routing between the remote node spawning this place and the place itself. The on-place-dead callback handles the event when the newly spawned place terminates.

method

(send a-place wait-for-die)  void?

Blocks and waits for the subprocess representing the remote-node% to exit.

class

connection% : class?

  superclass: object%

  extends: event-container<%>
The connection% instance represents a connection to a named-place instance running on the current node. It routes inter-node place messages to the named place.

constructor

(new connection%    
    [node node]    
    [name name]    
    [ch-id ch-id]    
    [sc sc])  (is-a?/c connection%)
  node : (is-a?/c remote-node%)
  name : string?
  ch-id : exact-positive-integer?
  sc : (is-a?/c socket-connection%)
Constructs a remote-connection% instance. The place-exec argument describes how the remote place should be launched in the same way as for launch-place in remote-node%. The ch-id and sc arguments are internally used to establish routing between the remote node and this named-place.

class

respawn-and-fire% : class?

  superclass: object%

  extends: event-container<%>
The respawn-and-fire% instance represents a thunk that should execute every n seconds.

constructor

(new respawn-and-fire% 
    [seconds seconds] 
    [thunk thunk]) 
  (is-a?/c respawn-and-fire%)
  seconds : (and/c real? (not/c negative?))
  thunk : (-> void?)
Constructs a respawn-and-fire% instance that when placed inside a message-router construct causes the supplied thunk to execute every n seconds.

class

after-seconds% : class?

  superclass: object%

  extends: event-container<%>
The after-seconds% instance represents a thunk that should execute after n seconds.

constructor

(new after-seconds% 
    [seconds seconds] 
    [thunk thunk]) 
  (is-a?/c after-seconds%)
  seconds : (and/c real? (not/c negative?))
  thunk : (-> void?)
Constructs an after-seconds% instance that when placed inside a message-router construct causes the supplied thunk to execute after n seconds.

class

restarter% : class?

  superclass: after-seconds%

  extends: event-container<%>
The restarter% instance represents a restart strategy.

constructor

(new restarter% 
    [seconds seconds] 
    [[retry retry] 
    [on-final-fail on-final-fail]]) 
  (is-a?/c restarter%)
  seconds : number?
  retry : (or/c number? #f) = #f
  on-final-fail : (or/c #f (-> any/c)) = #f
Constructs an restarter% instance that when supplied to a #:restart-on-exit argument, attempts to restart the process every seconds. The retry argument specifies how many time to attempt to restart the process before giving up. If the process stays alive for (* 2 seconds) the attempted retries count is reset to 0. The on-final-fail thunk is called when the number of retries is exceeded

value

distributed-launch-path : path?

Contains the local path to the distributed places launcher. The distributed places launcher is the bootsrap file that launches the message router on a new node.

procedure

(ssh-bin-path)  string?

Returns the path to the ssh binary on the local system in string form.

Example:

> (ssh-bin-path)

#<path:/usr/bin/ssh>

procedure

(racket-path)  path?

Returns the path to the currently executing Racket binary on the local system.

procedure

(build-distributed-launch-path collects-path)  string?

  collects-path : path-string?
Returns the path to the distributed places launch file. The function can take an optional argument specifying the path to the collects directory.

procedure

(spawn-node-at hostname 
  [#:listen-port port 
  #:racket-path racket-path 
  #:ssh-bin-path ssh-path 
  #:distributed-launch-path launcher-path]) 
  channel?
  hostname : string?
  port : port-no? = DEFAULT-ROUTER-PORT
  racket-path : string-path? = (racket-path)
  ssh-path : string-path? = (ssh-bin-path)
  launcher-path : string-path?
   = (path->string distributed-launch-path)
Spawns a node in the background using a Racket thread and returns a channel that becomes ready with a remote-node% once the node has spawned successfully

procedure

(spawn-nodes/join nodes-descs)  void?

  nodes-descs : list?
Spawns a list of nodes by calling (lambda (x) (apply keyword-apply spawn-node-at x)) for each node description in nodes-descs and then waits for each node to spawn.

procedure

(*channel-put ch msg)  void?

  ch : 
(or/c place-channel? async-bi-channel?
      channel? (is-a?/c remote-connection%))
  msg : any
Sends msg over ch channel.

procedure

(*channel-get ch)  any

  ch : 
(or/c place-channel? async-bi-channel?
      channel? (is-a?/c remote-connection%))
Returns a message received on ch channel.

procedure

(*channel? v)  boolean?

  v : any/c

procedure

(send-new-place-channel-to-named-dest ch 
  src-id 
  dest-list) 
  place-channel?
  ch : *channel?
  src-id : any
  dest-list : (listof string? port-no? string?)
Creates and returns a new place channel connection to a named place at dest-list. The dest-list argument is a list of a remote-hostname remote-port and named-place name. The channel ch should be a connection to a message-router.

procedure

(mr-spawn-remote-node mrch    
  host    
  [#:listen-port listen-port    
  #:solo solo])  void?
  mrch : *channel?
  host : string?
  listen-port : port-no? = DEFAULT-ROUTER-PORT
  solo : boolean? = #f
Sends a message to a message router over mrch channel asking the message router to spawn a new node at host listening on port listen-port. If the #:solo keyword argument is supplied the new node is not folded into the complete network with other nodes in the distributed system.

procedure

(mr-supervise-named-dynamic-place-at mrch    
  dest    
  name    
  path    
  func)  void?
  mrch : *channel?
  dest : (listof string? port-no?)
  name : string?
  path : string?
  func : symbol?
Sends a message to a message router over mrch channel asking the message router to spawn a named place at dest named name. The place is spawned at the remote node by calling dynamic place with module-path path and function func. The dest parameter should be a list of remote-hostname and remote-port.

procedure

(mr-connect-to mrch dest name)  void?

  mrch : *channel?
  dest : (listof string? port-no?)
  name : string?
Sends a message to a message router over mrch channel asking the message router to create a new connection to the named place named name at dest. The dest parameter should be a list of remote-hostname and remote-port.

procedure

(start-message-router/thread [#:listen-port listen-port 
  #:nodes nodes]) 
  
thread? channel?
  listen-port : port-no? = DEFAULT-ROUTER-PORT
  nodes : list? = null
Starts a message router in a Racket thread connected to nodes, listening on port listen-port, and returns a channel? connection to the message router.

procedure

(port-no? no)  boolean?

  no : (and/c exact-nonnegative-integer?  (integer-in 0 65535))
Returns #t if no is a exact-nonnegative-integer? between 0 and 65535.

The default port for distributed places message router.

class

named-place-typed-channel% : class?

  superclass: object%

The ch argument is a place-channel.

method

(send a-named-place-typed-channel get type)  any

  type : symbol?
Returns the first message received on ch that has the type type. Messages are lists and their type is the first item of the list which should be a symbol?. Messages of other types that are received are queued for later get requests.

procedure

(tc-get type ch)  void?

  type : symbol?
  ch : place-channel?
Gets a message of type type from the named-place-typed-channel% ch.

procedure

(write-flush datum port)  void?

  datum : any
  port : port?
Writes datum to port and then flushes port.

procedure

(printf/f format args ...)  void?

  format : string?
  args : any
Calls printf followed by a call to flush-output.

procedure

(displayln/f item)  void?

  item : any
Calls displayln followed by a call to flush-output.

Example:

> (write-flush "Hello World" (current-output-port))

"Hello World"

1 Define Remote Server

 (require racket/place/define-remote-server)
  package: distributed-places-lib

syntax

(define-remote-server [name identifier?] rpc-forms ...+)

syntax

(define-named-remote-server [name identifier?] rpc-forms ...+)

The define-remote-server and define-named-remote-server forms are nearly identical. The define-remote-server form should be used with supervise-dynamic-place-at to build a private rpc server, while the define-named-remote-server form should be used with supervise-named-dynamic-place-at to build a rpc server inside a named place.

The define-named-remote-server form takes an identifier and a list of custom expressions as its arguments. From the identifier a function is created by prepending the make- prefix. This procedure takes a single argument a place-channel. In the example below, the make-tuple-server identifier is the place-function-name given to the supervise-named-dynamic-place-at form to spawn an rpc server. The server created by the make-tuple-server procedure sits in a loop waiting for rpc requests from the define-rpc functions documented below.

syntax

(define-state id value)

Expands to a define, which is closed over by the define-rpc functions to form local state.

syntax

(define-rpc (id args ...) body ...)

Expands to a client rpc function name-id which sends id and args ... to the rpc server rpc-place and waits for a response. (define (name-id rpc-place args ...) body)

syntax

(define-cast (id args ...) body ...)

Expands to a client rpc function name-id which sends id and args ... to the rpc server rpc-place but does not receive any response. A cast is a one-way communication technique. (define (name-id rpc-place args ...) body)

The define-state custom form translates into a simple define form, which is closed over by the define-rpc forms.

The define-rpc form is expanded into two parts. The first part is the client stubs that call the rpc functions. The client function name is formed by concatenating the define-named-remote-server identifier, tuple-server, with the RPC function name set to form tuple-server-set. The RPC client functions take a destination argument which is a remote-connection% descriptor and then the RPC function arguments. The RPC client function sends the RPC function name, set, and the RPC arguments to the destination by calling an internal function named-place-channel-put. The RPC client then calls named-place-channel-get to wait for the RPC response.

The second expansion part of define-rpc is the server implementation of the RPC call. The server is implemented by a match expression inside the make-tuple-server function. The match clause for tuple-server-set matches on messages beginning with the 'set symbol. The server executes the RPC call with the communicated arguments and sends the result back to the RPC client.

The define-cast form is similar to the define-rpc form except there is no reply message from the server to client

Example:

> (module tuple-server-example racket/base
    (require racket/match
             racket/place/define-remote-server)
  
    (define-named-remote-server tuple-server
      (define-state h (make-hash))
      (define-rpc (set k v)
        (hash-set! h k v)
        v)
      (define-rpc (get k)
        (hash-ref h k #f))
      (define-cast (hello)
        (printf "Hello from define-cast\n")
        (flush-output))))

Example:

> (module bank-server-example racket/base
    (require racket/match
             racket/place/define-remote-server)
  
    (define-remote-server bank
      (define-state accounts (make-hash))
      (define-rpc (new-account who)
         (match (hash-has-key? accounts who)
           [#t '(already-exists)]
           [else
             (hash-set! accounts who 0)
             (list 'created who)]))
      (define-rpc (remove who amount)
         (cond
           [(hash-ref accounts who (lambda () #f)) =>
              (lambda (balance)
                (cond [(<= amount balance)
                       (define new-balance (- balance amount))
                       (hash-set! accounts who new-balance)
                       (list 'ok new-balance)]
                      [else
                        (list 'insufficient-funds balance)]))]
           [else
             (list 'invalid-account who)]))
      (define-rpc (add who amount)
        (cond
           [(hash-ref accounts who (lambda () #f)) =>
              (lambda (balance)
                (define new-balance (+ balance amount))
                (hash-set! accounts who new-balance)
                (list 'ok new-balance))]
           [else
             (list 'invalid-account who)]))))

procedure

(log-to-parent msg [#:severity severity])  void?

  msg : string?
  severity : symbol? = 'info
The log-to-parent procedure can be used inside a define-remote-server or define-named-remote-server form to send a logging message to the remote owner of the rpc server.

2 Async Bidirectional Channels

 (require racket/place/private/async-bi-channel)
  package: base

Creates and returns an opaque structure, which is the async bidirectional channel.

procedure

(async-bi-channel? ch)  boolean?

  ch : any
A predicate that returns #t for async bidirectional channels.

procedure

(async-bi-channel-get ch)  any

  ch : async-bi-channel?
Returns the next available message from the async bidirectional channel ch.

procedure

(async-bi-channel-put ch msg)  void?

  ch : async-bi-channel?
  msg : any
Sends message msg to the remote end of the async bidirectional channel ch.

3 Distributed Places MPI

 (require racket/place/distributed/rmpi)
  package: distributed-places-lib

struct

(struct rmpi-comm (id cnt channels))

  id : exact-nonnegative-integer?
  cnt : exact-positive-integer?
  channels : vector?
The communicator struct rmpi-comm contains the rmpi process rank id, the quantity of processes in the communicator group, cnt, and a vector of place-channels, one for each process in the group.

procedure

(rmpi-id comm)  exact-nonnegative-integer?

  comm : rmpi-comm?
Takes a rmpi communicator structure, comm, and returns the node id of the RMPI process.

procedure

(rmpi-cnt comm)  exact-positive-integer?

  comm : rmpi-comm?
Takes a rmpi communicator structure, comm, and returns the count of the RMPI processes in the communicator group.

procedure

(rmpi-send comm dest val)  void?

  comm : rmpi-comm?
  dest : exact-nonnegative-integer?
  val : any
Sends val to destination rmpi process number dest using the RMPI communicator structure comm.

procedure

(rmpi-recv comm src)  any

  comm : rmpi-comm?
  src : exact-nonnegative-integer?
Receives a message from source rmpi process number src using the RMPI communicator structure comm.

Creates the rmpi-comm structure instance using the named place’s original place-channel ch. In addition to the communicator structure, rmpi-init returns a list of initial arguments and the original place-channel ch wrapped in a named-place-typed-channel%. The named-place-typed-channel% wrapper allows for the reception of list messages typed by an initial symbol.

procedure

(rmpi-broadcast comm src)  any

  comm : rmpi-comm?
  src : exact-nonnegative-integer?
(rmpi-broadcast comm src val)  any
  comm : rmpi-comm?
  src : exact-nonnegative-integer?
  val : any
Broadcasts val from src to all rmpi processes in the communication group using a hypercube algorithm. Receiving processes call (rmpi-broadcast comm src).

procedure

(rmpi-reduce comm dest op val)  any

  comm : rmpi-comm?
  dest : exact-nonnegative-integer?
  op : procedure?
  val : any
Reduces val using the op operator to dest rmpi node using a hypercube algorithm.

procedure

(rmpi-barrier comm)  void?

  comm : rmpi-comm?
Introduces a synchronization barrier for all rmpi processes in the communcication group comm.

procedure

(rmpi-allreduce comm op val)  any

  comm : rmpi-comm?
  op : procedure?
  val : any
Reduces val using the op operator to rmpi node 0 and then broadcasts the reduced value to all nodes in the communication group.

Partitions num into rmpi-cnt equal pieces and returns the offset and length for the rmpi-idth piece.

procedure

(rmpi-build-default-config 
  #:racket-path racket-path 
  #:distributed-launch-path distributed-launch-path 
  #:mpi-module mpi-module 
  #:mpi-func mpi-func 
  #:mpi-args mpi-args) 
  hash?
  racket-path : string?
  distributed-launch-path : string?
  mpi-module : string?
  mpi-func : symbol?
  mpi-args : (listof any)
Builds a hash from keywords to keyword arguments for use with the rmpi-launch function.

procedure

(rmpi-launch default-node-config config)  void?

  default-node-config : hash?
  config : 
(listof (list/c string? port-no? symbol?
                exact-nonnegative-integer?))
Launches distributed places nodes running #:mpi-func in #:mpi-module with #:mpi-args. The config is a list of node configs, where each node config consists of a hostname, port, named place symbol and rmpi id number, followed by and optional hash of keyword #:racket-path, #:distributed-launch-path, #:mpi-module, #:mpi-func, and #:mpi-args to keyword arguments. Missing optional keyword arguments will be taken from the default-node-config hash of keyword arguments.

procedure

(rmpi-finish comm tc)  void?

  comm : rmpi-comm?
  tc : (is-a?/c named-place-typed-channel%)
Rendezvous with the rmpi-launch, using the tc returned by rmpi-launch, to indicate that the RMPI module is done executing and that rmpi-launch can return control to its caller.

Example:

> (rmpi-launch
      (rmpi-build-default-config
        #:racket-path "/tmp/mplt/bin/racket"
        #:distributed-launch-path (build-distributed-launch-path
                                   "/tmp/mplt/collects")
        #:mpi-module "/tmp/mplt/kmeans.rkt"
        #:mpi-func   'kmeans-place
        #:mpi-args    (list "/tmp/mplt/color100.bin" #t 100
                            9 10 1e-07))
  
      (list (list "nodea.example.com" 6340 'kmeans_0 0)
            (list "nodeb.example.com" 6340 'kmeans_1 1)
            (list "nodec.example.com" 6340 'kmeans_2 2)
            (list "noded.example.com" 6340 'kmeans_3 3)))